Computing systems with direct attached storage (DAS) generally require involvement of the processor in order to perform snapshots and data backups, since it is only through the processor that the direct attached storage is accessed. Snapshots and data backups thus disrupt execution of applications and may require application downtime or scheduling. Snapshots occupy storage space on a primary compute node where applications are running, resulting in extra space requirements on the compute node. Restoring data to direct attached storage in such a computing system also disrupts execution of applications, and may be impossible when an application or processor has crashed. Also, in systems with direct attached storage, backups, antivirus scans, and extraction transformation and loading (ETL) processing require use of resources on a primary compute node, which may slow execution of applications. Therefore, there is a need in the art for a solution which overcomes the drawbacks described above.